On-Call Handoffs with AI agents

A Neotask agent runs the on-call handoff by pulling the current PagerDuty schedule, summarizing every incident that opened during the outgoing shift, and posting a structured handoff note to the team's Slack channel before the new on-call engineer's shift starts — so the incoming engineer starts their shift already knowing what is in flight instead of reconstructing it from scrollback.

How it works today vs. with Neotask

Incident context dies in the gap between shifts more often than anywhere else in an on-call rotation. The outgoing engineer knows exactly which alert is a known flake, which incident is still being actively worked, and which ticket needs a follow-up in the morning — but that knowledge lives in their head, not in a system, and if they do not personally write it down before logging off, it is gone. The incoming engineer then either pings the outgoing one at an inconvenient hour or spends the first thirty minutes of their shift re-deriving context that already existed twelve hours earlier. A standardized, automatically generated handoff removes the dependency on the outgoing engineer remembering to write good notes at the exact moment they are trying to log off.

The agent flow

Pull the current and upcoming on-call schedule

The agent reads the active PagerDuty schedule to confirm who is finishing their shift and who is starting the next one, timing the handoff to run ahead of the transition.

Integration: pagerduty

Summarize open incidents

Every incident still open from the outgoing shift is pulled from PagerDuty with its current status, severity, and last update, rather than relying on the outgoing engineer to remember which ones are unresolved.

Integration: pagerduty

Cross-reference linked tickets

Any Jira tickets tied to those incidents are checked for status, so the handoff note reflects whether follow-up work has already been filed or still needs to be.

Integration: jira

Draft the structured handoff note

A consistent format — open incidents, known issues, anything to watch, deploys in progress — is generated instead of a free-text message that varies by whoever is writing it.

Post to the on-call Slack channel

The handoff note posts to the team's dedicated on-call channel at shift change, tagging the incoming engineer directly so it is not missed among other channel activity.

Integration: slack

Update the shared runbook if needed

If a new failure mode was discovered during the shift, the agent drafts a runbook addition in Notion for the outgoing engineer to confirm rather than letting the knowledge stay tribal.

Integration: notion

Variations

• Teams running multiple services stagger handoffs per service rotation instead of a single company-wide handoff message.
• For a quiet shift with zero incidents, the agent posts an abbreviated note confirming clean handoff rather than the full structured format, keeping channel noise proportional to what happened.
• Some teams require the incoming engineer to react-confirm the handoff message in Slack before the outgoing engineer is considered off-call.

Frequently asked questions

Does this replace a live handoff conversation?

For most shift changes it does; teams can still hold a live call for particularly complex ongoing incidents, using the written note as the shared reference point.

How does it handle incidents that started right at shift change?

It timestamps the handoff generation and captures whatever incident state exists at that moment, then updates the note if status changes shortly after.

What if PagerDuty and Jira disagree on an incident's status?

The mismatch is flagged explicitly in the handoff note rather than silently trusting one system, since that disagreement usually means someone forgot to update a ticket.

Can it include metrics or dashboards?

Yes, teams can configure it to link relevant Datadog or monitoring dashboards directly in the handoff note for anything flagged as still elevated.

Plans

Continue